Feedback on new Continuous Space

[EwoutH]

Hi everyone! We've just released a significant update to Mesa's continuous space functionality. This reimplementation introduces an agent-centric API, n-dimensional space support, and various performance improvements. We'd love to hear your feedback, experiences, and suggestions.

Key Changes

The new implementation features:

• An agent-centric API similar to cell spaces (e.g., agent.position property)
• Direct vector arithmetic for movement (agent.position += [0.1, 0.1])
• Efficient neighbor calculations using optimized distance computations
• Support for n-dimensional spaces
• Better memory management with dynamic array resizing

Here's an overview of the new interface:

# Create a 2D continuous space
space = ContinuousSpace(
    dimensions=[[0, 1], [0, 1]],
    torus=True,
    random=model.random
)

# Create and position an agent
agent = ContinuousSpaceAgent(space, model)
agent.position = [0.5, 0.5]

# Move agent using vector arithmetic
agent.position += [0.1, 0.1]

# Get neighbors within radius
neighbors, distances = agent.get_neighbors_in_radius(radius=0.2)

# Find k nearest neighbors
nearest, distances = agent.get_nearest_neighbors(k=5)

And if you want to migrate an existing model, here's a quick before/after comparison:

# Previous Implementation
space = ContinuousSpace(x_max=1.0, y_max=1.0, torus=True)
agent = MyAgent(unique_id=1, model=model)
space.place_agent(agent, (0.5, 0.5))
space.move_agent(agent, (0.6, 0.6))

# New Implementation 
space = ContinuousSpace(dimensions=[[0, 1], [0, 1]], torus=True, random=model.random)
agent = ContinuousSpaceAgent(space, model)
agent.position = [0.5, 0.5]
agent.position += [0.1, 0.1]

For more details, see:

• the PR: Reimplementation of Continuous Space #2584
• the API documentation

Discussion Points

We're particularly interested in hearing about:

1. How does the new API feel in practice? Is it more intuitive?
2. Have you noticed any performance differences in your models?
3. Are you using or planning to use the n-dimensional space support?
4. Any challenges in migrating existing models?
5. Features you'd like to see added or improved?

Please share your thoughts, questions, and any issues you encounter!